Data Center Revenue: Nvidia Made More in One Quarter Than AMD, Intel, and IBM in a Year

Data centers are the defining infrastructure investment of the artificial intelligence (AI) era. The largest technology companies are spending at levels that far exceed any previous build-out in the industry’s history, and that spending flows directly into the revenue of AI companies that supply chips, servers, and cloud capacity.

The results have not been equally distributed. Nvidia (NVDA -1.56%) has captured the dominant share of AI chip spending. Broadcom (AVGO -4.11%) has built the second-largest AI semiconductor business through custom chip deals. Intel (INTC +1.14%) is fighting to recover lost market share. Cloud operators running the infrastructure are planning capital expenditures of hundreds of billions of dollars in 2026 alone, and they might be coming under pressure to start showing a return on that investment.

How revenue has stacked up between and among hardware suppliers and cloud operators could offer investors clues into which industries and companies are poised for growth.

How Do AI Companies Make Revenue From Data Centers?

Data center revenue can be measured in two groups of companies:

• AI chip and hardware suppliers, which make and sell the equipment that data centers run on.
• Cloud operators, which build and operate data centers and sell access to computing capacity.

Their revenue models are distinct, and the figures aren’t directly comparable.

Revenue reflects each company's reported data center or AI hardware segment. Nvidia and Broadcom figures are on different fiscal-year calendars than the others. Cloud services revenue reflects fees charged to businesses for computing capacity. This is not directly comparable to chip segment revenue. Sources: company earnings releases.

AI Chip and Hardware Suppliers: Revenue by Company

Five companies account for the bulk of AI chip and data center hardware revenue. Nvidia sits well above the rest. The others are competing for the remaining share of the fast-growing market.

Cloud Operator Data Center Revenue: AWS, Azure, and Google Cloud

Chip suppliers sell to the data center market, which are driven by hyperscalers. Amazon (AMZN -0.87%) Web Services (AWS), Microsoft (MSFT -1.57%) Azure, and [Alphabet’s (GOOG -0.58%)] Google Cloud are the largest buyers of AI chips and the primary providers of AI computing capacity.

Their AI data center cloud revenue is not directly comparable to data center revenue from chip companies – it reflects what businesses pay to use infrastructure, not the cost to build it. Both figures describe revenue measured at different points of the AI supply chain.

What Is Next for Data Center Revenue?

Forward guidance points to continued growth in data center revenue. For example:

• Nvidia guided for total revenue of $78 billion for Q1 FY2027.
• Broadcom expects AI chip sales to approximately double year over year in the current quarter.
• AMD projects more than 60% annual data center segment growth over the next several years.
• The hyperscalers are collectively planning close to $600 billion in 2026 capital expenditure.

The biggest structural question for chip suppliers is whether hyperscalers will eventually build enough of their own silicon to reduce third-party GPU demand. Google has been developing its own Tensor Processing Units (TPUs) since 2016. Amazon has its Trainium training chip and Inferentia inference chip. Microsoft is developing its own AI accelerator. The incentive for hyperscalers is clear: reduce dependence on a single supplier and lower long-term costs by taking chip design in-house.

The constraint is equally clear: Nvidia's CUDA software platform has a decade-long head start, and most AI developers build on it by default. Even hyperscalers that have custom chips in production still buy large quantities of Nvidia GPUs to serve customers locked into CUDA-based workflows.

There are other risks to AI data center revenue worth tracking as well:

• AI model efficiency: Models are becoming more capable per dollar of compute. If efficiency gains outpace demand growth, infrastructure spending could moderate. Historically, cheaper compute has expanded total usage rather than replacing it, but that is not guaranteed.
• Export controls: U.S. restrictions on chip sales to China are already creating revenue headwinds for Nvidia and AMD.
• Capital expenditure monetization: Hyperscalers are spending at historically high levels relative to operating cash flows. If AI services revenue does not grow fast enough to justify the investment, capital expenditure plans could be revised downward, reducing demand for chip suppliers.

FAQs